The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python script (scripts/inspect_media.py) that executes the ffprobe command to extract media metadata. The execution is performed using subprocess.run with a structured list of arguments and without a shell, which is a secure implementation that prevents command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes external media files through its inspection utility. While this technically creates an indirect prompt injection surface, the risk is mitigated because the script selectively extracts only technical metadata (such as resolution and frame rate) and ignores potentially untrusted string-based metadata tags.
[SAFE]: The implementation of the React component and the accompanying documentation follow standard web development practices, including secure canvas pixel sampling and appropriate performance safeguards.

video-to-particle-field